455kHz bandpass filter for manufacture.

[arildj78]

What would be a good way of designing a 455kHz bandpass filter for manufacture? I've looked at elliptic filter calculators and Analog Devices filter wizard with their multiple feedback active filter. The passband should be as narrow as possible, preferable down to a few hundred Hz, but I'll take whatever I can get.

As far as I can tell I end up with a lot of items on my BOM, just for this filter. Also, the tolerances are killing me! Even with 1% parts, the gain is all over the shop.

A ceramic filter would be nice but apparently they are all end-of-life so I'm not designing in one of those.

Keep in mind that design for manufacture is an important aspect on this one...  Anyone got a suggestion?

[Miti]

Did you look at Murata?

https://www.ttiinc.com/content/ttiinc/en/manufacturers/k-o/murata/products/murata_piezoelectric_ceramic_filters.html

[Whales]

Another possibility:double the freq and use a 1MHz crystal?

[arildj78]

Murata is at end of life on their 455kHz filters. It looks like this is ancient tech, that either no one is using, or is replaced by something else?

[arildj78]

I could double the freq, but my RF is at this frequency, so that would mean adding a LO, mixer and whole lot of stuff...

[ezalys]

Switched capacitor may be the move.

[T3sl4co1l]

The modern way would be DSP...

Tim

[arildj78]

How would I even get started with that DSP? I'm currently using a SAMD21 (Cortex M0+) which is probably to lightweight to pull off DSP. Feel free to point me in a general direction both for processor and for literature.

[ogden]

"For manufacture" means significant volume, right? In such case you ignore what distributors say about product lifetime, contact sales of manufacturer directly and negotiate. DSP approach would be more or less building simple software defined radio (receiver). As there is more than one "DSP approach", would be nice to know - what is your application, what are signal properties & source and what you want to do with filtered signal in the end?

[T3sl4co1l]

Hmmm... not sure exactly how much CPU power you need.  A more manageable analog filter might have a bandwidth of 10-20%, in which case you need a sample rate only modestly higher than the center frequency itself.  For which a SAMD21 might actually have enough sample rate out of the box, but you may consider an external ADC for better performance as well.

You'd then need, let's see... you'll want a digital filter to start off, with modest bandwidth, to clean up the band skirts left from the analog filter; this might be a couple IIR filters needing say 4-16 MAC operations.  You want to downconvert (decimate or mix) at some point, which should take a few more MACs and such.  I don't know the exact performance of the SAMD21 but typical ARMs at 100MHz or thereabouts ought to be enough for that, or a little marginal, in which case a more DSP-tuned architecture, or higher clock, would do.

Once it's converted down, you can demodulate directly (direct conversion, implemented digitally), or keep it at a modest center frequency and continue filtering and stuff.  Eventually I guess you'll be demodulating it in whatever way; if that requires a 90 degree phase shift (Hilbert transform), or FFT, you'll probably want something beefier.

You will want to have the ADC output buffered I think, so the DSP operations can be done in blocks.  Interrupt per sample might not be too bad, but it does bring in a fair amount of overhead, and you don't have too much time to spend before the next sample comes in.

If you can reduce the bandwidth at any earlier point, you can of course get by with less CPU power.  A signal of just a few kHz bandwidth could even be handled by an Arduino or PIC, a Cortex M0+ is definitely overkill compared to the minimum possible processing (but, that would require a faster sample-and-hold than an Arduino's ADC can do, and a much tighter analog filter which doesn't save you from the OP problem ).

I'm not an expert on digital radio systems, and it may be there are better optimizations.  If nothing else, you can write out the signal chain in exactly the same way you'd solve it analog, then implement the difference functions to solve them in DSP.  That's kind of where I'm approaching this from.

Tim

[arildj78]

I'm measuring the signal strength of a 455kHz CW pulse. The pulse is 180ms long and repeats at about 1Hz. I'm using an RSSI chip to do the detection, so all I need to do is to filter all unwanted signals.

[arildj78]

freq is 455, required BW 200Hz.

[David Hess]

Murata is at end of life on their 455kHz filters. It looks like this is ancient tech, that either no one is using, or is replaced by something else?

The IF filters required by superhetrodyne receivers cannot be integrated so poorer performing direct conversion receivers have replaced them.

[ogden]

I'm measuring the signal strength of a 455kHz CW pulse. The pulse is 180ms long and repeats at about 1Hz. I'm using an RSSI chip to do the detection, so all I need to do is to filter all unwanted signals.

In such case I would use simplest solution I can imagine: low band-pass "preselector" LC filter (3rd, max 5th order),  synchronous demodulator made out of analog mux. Filtered baseband you slowly sample with ADC (2Ksps or so), then do envelope detector in software. More about sync demodulation here

[vk4ffab]

Inrad still make 455khz narrow filters, but you will be paying in kidneys for them. You cant up convert to say 9mhz and build a narrow xtal filter. All current 455khz filters i know of are for am and are 6khz wide, not 200hz.

[fcb]

As others have said. 200Hz is to small a BW for convential old school 455KHz IF filters.

I would be looking at either undersampling (see sampling scopes) with your microcontroller ADC, or mixing it down to baseband and sampling, or perhaps switched capacitor bandpass filter (possibly a bit high for this).

Does your signal vary in amplitude over the 180ms window, are you looking for the average over that time?  Required accuracy? What do your interfering signals look like and what level?  Do you have a budget?

You could potentially PLL to the incoming signal and use a lock-in amplifier technique to extract magnitude.

[eliocor]

Maybe this can be of some interest to you: https://www.i2phd.org/armradio/index.html

[BigBoss]

It's worth to take a glance at
Murata CFJ455K8
https://www.ebay.fr/itm/Murata-CFJ455K8-455-KHz-High-Performance-Ceramic-Filter-for-Radio-Equipment-/352717790445

[David Hess]

If you want a narrow 455kHz IF filter that you can buy immediately, you might look for what is available from the manufacturers of amateur radio HF transceivers.  It is a common user installable option.

[vk4ffab]

If you want a narrow 455kHz IF filter that you can buy immediately, you might look for what is available from the manufacturers of amateur radio HF transceivers.  It is a common user installable option.

They are all pretty much out of production. Inrad is the only one still with production, but for how long is anyones guess. And they are dear as poison. When your b9m starts with 250 bucks for a filter, change your design.

[David Hess]

If you want a narrow 455kHz IF filter that you can buy immediately, you might look for what is available from the manufacturers of amateur radio HF transceivers.  It is a common user installable option.

They are all pretty much out of production. Inrad is the only one still with production, but for how long is anyones guess. And they are dear as poison. When your b9m starts with 250 bucks for a filter, change your design.

What I am getting at is that they can be bought as a module directly from Kenwood, Yaesu, Icom, whatever, or from Ebay.  Do a search through their HF radios to see which have narrow 455kHz IF filters as an option.

Obviously this would only work for a small number of units and the cost is high.

For 455kHz, the old school way using a pair or more of slug tuned IF transformers and a variable coupling capacitor might be the best at this point but it requires tuning.  Nothing prevents doing this exclusively with trimmer capacitors and fixed tapped inductors.

[KaneTW]

Is anything stopping you from doing this in DSP? I'm in a similar boat for a 20MHz video FM receiver with 1MHz deviation and after some fiddling around just decided to do it in DSP.

[vk4ffab]

If you want a narrow 455kHz IF filter that you can buy immediately, you might look for what is available from the manufacturers of amateur radio HF transceivers.  It is a common user installable option.

They are all pretty much out of production. Inrad is the only one still with production, but for how long is anyones guess. And they are dear as poison. When your b9m starts with 250 bucks for a filter, change your design.

What I am getting at is that they can be bought as a module directly from Kenwood, Yaesu, Icom, whatever, or from Ebay.  Do a search through their HF radios to see which have narrow 455kHz IF filters as an option.

Obviously this would only work for a small number of units and the cost is high.

For 455kHz, the old school way using a pair or more of slug tuned IF transformers and a variable coupling capacitor might be the best at this point but it requires tuning.  Nothing prevents doing this exclusively with trimmer capacitors and fixed tapped inductors.

Yeah the caveat here is he said its for a commercial product. If your design starts with a $250 part, it better have super high margin, if your design starts with end of life and surplus, you better change your design. There are dozens of ways to get single signal reception that do not require the use of stupid expensive or hard to get parts. Just change the IF frequency to one where you can build a suitable narrow filter or do away with analog entirely and go DSP. 

[David Hess]

Yeah the caveat here is he said its for a commercial product. If your design starts with a $250 part, it better have super high margin, if your design starts with end of life and surplus, you better change your design. There are dozens of ways to get single signal reception that do not require the use of stupid expensive or hard to get parts. Just change the IF frequency to one where you can build a suitable narrow filter or do away with analog entirely and go DSP.

IF filters are awfully useful to improve in-band rejection of signals which would otherwise overload the following stage.  No amount of DSP can undue overload of the final mixer or digitizer.

As I recall, Rockwell bought Collins' mechanical filter division for exactly this reason and continued to sell mechanical filters to others but of course later shut it down.

If the IF frequency can be lowered, then bandpass switched capacitor filters become viable.

[ogden]

IF filters are awfully useful to improve in-band rejection of signals which would otherwise overload the following stage.  No amount of DSP can undue overload of the final mixer or digitizer.

OP did not even mention specifics of interference, yet you are suggesting most advanced and expensive solution one can imagine. HAM HF rig IF filters are as obsolete as Murata 455KHz ceramic filters, they cost ~ 100x more. I find it inappropriate for knowledgeable engineer to suggest something like that - for application "measuring the signal strength of a 455kHz CW 180ms pulse".

[vk4ffab]

Yeah the caveat here is he said its for a commercial product. If your design starts with a $250 part, it better have super high margin, if your design starts with end of life and surplus, you better change your design. There are dozens of ways to get single signal reception that do not require the use of stupid expensive or hard to get parts. Just change the IF frequency to one where you can build a suitable narrow filter or do away with analog entirely and go DSP.

IF filters are awfully useful to improve in-band rejection of signals which would otherwise overload the following stage.  No amount of DSP can undue overload of the final mixer or digitizer.

As I recall, Rockwell bought Collins' mechanical filter division for exactly this reason and continued to sell mechanical filters to others but of course later shut it down.

If the IF frequency can be lowered, then bandpass switched capacitor filters become viable.

If you go DSP you do not need an IF at all and thus the IF filter. Just direct convert to audio I and Q and then do the processing in software with Hilbert transform and FIR filters. For the cost of a $2 Arm micro a couple of mixers and quadrature LO from an SI5351A. The whole receiver build would be sub $50. There are plenty of examples of this on the net from Charlie Morris, Pete Juliano etc and this is the way the QSX from QRPlabs, which is being designed the exact same way.

[arildj78]

Inrad still make 455khz narrow filters, but you will be paying in kidneys for them. You cant up convert to say 9mhz and build a narrow xtal filter. All current 455khz filters i know of are for am and are 6khz wide, not 200hz.

The narrower the better, but I'd take 6kHz any day.

[arildj78]

Inrad still make 455khz narrow filters, but you will be paying in kidneys for them. You cant up convert to say 9mhz and build a narrow xtal filter. All current 455khz filters i know of are for am and are 6khz wide, not 200hz.

I just looked at the Inrad filters and you're right, had to pay one kidney, my right arm and my first born child for that thing! Also, they were out-of-stock   

[arildj78]

Is anything stopping you from doing this in DSP? I'm in a similar boat for a 20MHz video FM receiver with 1MHz deviation and after some fiddling around just decided to do it in DSP.

This is a sidegig, and my knowledge is not up to the task yet. In a future implementation I hope to do this with a micro, but for now, I figured that the easiest path would be through the analog domain.

Little did I know... If there are any greybeards out there in search of a Norwegian apprentice, feel free to give me a shout

[fcb]

If you just want to get something up & running quickly, then you can build an 8th order bandpass (20KHz BW) using easily available parts:
https://www.analog.com/designtools/en/filterwizard/

[arildj78]

If you just want to get something up & running quickly, then you can build an 8th order bandpass (20KHz BW) using easily available parts:
https://www.analog.com/designtools/en/filterwizard/

This is the approach I've started on. Do you have any suggestion on how to minimize the effect of component tolerance? In my first try, both the center freq and the gain has a very large tolerance envelope.

[KaneTW]

Use 0.1% tolerance components and it should be well within margin. Note that narrow (high-Q) active bandpass filters very quickly get unrealistically high GBW requirements. GBW requirements for an active filter are proportional to Q * gain * frequency, multiplied by some safety factor to account for nonideal opamps (https://www.ti.com/lit/an/sbaa236/sbaa236.pdf, https://www.changpuak.ch/electronics/downloads/sloa088.pdf).

Traditionally the safety factor is 100 for a max. 1% gain error, but I find that way too high for most applications.

Component nonideality is another good reason to use digital filters. A FIR filter isn't that difficult to design and simulate (I use ScopeFIR for figuring out the frequency response and tap count, then Mathematica for the actual calculations and simulations).

[vk4ffab]

Is anything stopping you from doing this in DSP? I'm in a similar boat for a 20MHz video FM receiver with 1MHz deviation and after some fiddling around just decided to do it in DSP.

This is a sidegig, and my knowledge is not up to the task yet. In a future implementation I hope to do this with a micro, but for now, I figured that the easiest path would be through the analog domain.

Little did I know... If there are any greybeards out there in search of a Norwegian apprentice, feel free to give me a shout

It might be possible to use a high speed quad opamp and do an 8th order high pass followed by an 8th order low pass and get a fairly narrow bandpass that way shoot for say 1K, this will give much better skirts and slope  factor to the filter compared to using the bandpass calculator values. You could try simulating this in LTspice to see if you can get  good enough shape.

But there are probably easier ways to go about this, but its hard to give specifics without knowing much about what you are trying to achieve. As for DSP, its not all that difficult, if you want to go down that rabbit hole, here is a series of 10 vids by Charlie Morris ZL2CTM in New Zealand who has done most of the work for us. Basically you can do it all with a TeensyDuino and a handful of parts.



Rob.

[David Hess]

It might be possible to use a high speed quad opamp and do an 8th order high pass followed by an 8th order low pass and get a fairly narrow bandpass that way shoot for say 1K, this will give much better skirts and slope  factor to the filter compared to using the bandpass calculator values. You could try simulating this in LTspice to see if you can get  good enough shape.

At least one Linear Technology application note briefly discusses using current feedback amplifiers as part of an IF filter circuit but it was not an active filter.  Before fast operational amplifiers became available, sometimes discrete transistors were used for high frequency active filters.  I think the complexity and tolerance requirements would get out of control very quickly when trying to achieve the performance of a resonator but maybe it would be good enough.

[fmkit]

I  just tested simple 2 series 455kHz ceramic resonators filter ( as normally used crystals),   these marked 455E.
Bandpass CF: 435.45kHz,   Bandwidth 2kHz
Sweep 100kHz, center 455 

ps: added sweep of  LT455HW  ceramic filter,  measured BW is 10.4kHz,  2-series res.filter  5 times narrower and much cheaper,  I'll use one on DIY spectrum analyzer to see FM broadcast 19kHz pilot